Research Report

At first sight: how do restrained eaters evaluate

high-fat palatable foods?

A. Roefsa,*, C.P. Hermanb, C.M. MacLeodc, F.T.Y. Smuldersa, A. Jansena

aDepartment of Experimental Psychology, Faculty of Psychology (UNS 40), Maastricht University,

P.O. Box 616, 6200 MD Maastricht, The NetherlandsbDepartment of Psychology, University of Toronto, 100 St. George Street, Toronto, Ontario, Canada M5S 3G3

cDepartment of Psychology, University of Waterloo, 200 University Avenue West, Waterloo, Ontario, Canada N2L 3G1

Received 24 May 2004; revised 6 July 2004; accepted 2 August 2004

Abstract

Two experiments tested the hypothesis that restrained eaters display a greater liking for high-fat palatable foods, than do unrestrained

eaters. This hypothesis was tested in the affective priming paradigm [Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R.

(1986). On the automatic activation of attitudes. Journal of Personality and Social Psychology, 50, 229–238] and in the extrinsic affective

Simon task [De Houwer, J. (2003a). The extrinsic affective Simon task. Experimental Psychology, 50, 77–85]. Both paradigms were

successful in uncovering food likes and dislikes, and both showed that participants were able to evaluate the palatability of foods relatively

automatically. However, contrary to the hypothesis, food likes were not substantially affected by fat content, nor were they affected by

restraint-status. Restrained and unrestrained eaters may like high-fat palatable foods to the same extent, but may differ in their craving for

these foods.

q 2004 Elsevier Ltd. All rights reserved.

Keywords: Restrained eaters; Food likes; Indirect measures; Affective priming paradigm; Extrinsic affective Simon task

Introduction

What determines which foods people choose to

consume? This question is difficult to answer given the

many factors that could influence food choice (Mela,

1999, 2001), factors such as palatability, health concerns,

and availability. One of the most important of these

factors seems to be the palatability (liking) of the food

(Eertmans, Baeyens, & Van den Bergh, 2001), a factor

that affects almost everyone. Moreover, there may also

exist individual differences in the liking of foods,

governed by such characteristics as one’s body weight

and the extent to which one desires to control that

weight. In the current research, we are interested in

0195-6663/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.appet.2004.08.001

* Corresponding author.

E-mail address: a.roefs@psychology.unimaas.nl (A. Roefs).

precisely those individual differences in food liking. The

first question that arises is how these differences can best

be measured.

Some studies (Gerding & Weinstein, 1992) have been

conducted in which participants were simply asked to report

what kinds of food they like. An obvious disadvantage of

this kind of measure is that socially desirable answering

tendencies can be fairly strong. It might not be easy to admit

a liking for high-fat foods in a society in which being slim is

considered very important, and in which obesity is

stigmatized (Puhl & Brownell, 2003; Teachman & Brow-

nell, 2001). To try to overcome the disadvantages of

questionnaires, Lamote, Hermans, Baeyens, and Eelen

(2004) used an indirect measure to assess food likes and

dislikes. More specifically, they used the affective priming

paradigm (Fazio, Sanbonmatsu, Powell, & Kardes, 1986) to

study food likes and dislikes in a healthy population.

Appetite 44 (2005) 103–114

www.elsevier.com/locate/appet

A. Roefs et al. / Appetite 44 (2005) 103–114104

The results of their study suggest that the affective priming

paradigm is a suitable measure to uncover people’s food

likes and dislikes.

The affective priming paradigm is one of the indirect

measures that have recently become very popular in clinical

psychology (Palfai & Ostafin, 2003; Sherman, Rose, Koch,

Presoon, & Chassin, 2003) and social psychology (Dovidio,

Kawakami, & Gaertner, 2002; Fazio, Jackson, Dunton, &

Williams, 1995). In this paradigm (Fazio et al., 1986; Klauer

& Musch, 2003) two stimuli are presented in quick

succession, a prime followed by a target. No response is

required to the prime, which is simply displayed and

replaced by the target. Participants have to respond to the

target by evaluating it in terms of it is associated with

positive versus negative affect. The dependent variable is

the positive/negative key-press latency in response to the

target.

The focus of the priming paradigm is on the extent to

which the presentation of the prime influences the response

to the target. Typically (Bargh, Chaiken, Govender, &

Pratto, 1992; Fazio et al., 1986; Hermans, De Houwer, &

Eelen, 1994, 2001), affectively congruent prime–target

pairs (e.g. ‘love’—‘happy’) lead to shorter response

latencies to the target word than do affectively incongruent

prime–target pairs (e.g. love—‘awful’). This in itself is not

surprising, and is consistent with semantic priming effects

more generally (see Neely, 1991). The critical idea is that

the pattern of response latencies as a function of affect

match between prime and target indicates how people

evaluate the prime on a fairly automatic level. Applied to

the palatability of food, if people respond faster on

congruent trials (‘palatable—positive’ and ‘unpalatable—

negative’) than on incongruent trials (‘palatable—negative’

and ‘unpalatable—positive’), it can be inferred that they

like palatable foods more than unpalatable foods. The main

advantage of employing such indirect measures is that they

can estimate people’s evaluations of various stimuli

without directly asking (Fazio & Olson, 2003), thereby

possibly reducing the risk of socially desirable answering

tendencies. Responses are assumed to be relatively

automatic in this kind of task, because stimuli are

presented in quick succession and participants are urged

to respond as quickly as possible, leaving insufficient time

for controlled processing1 (Hermans, De Houwer et al.,

2001).

1 Using an indirect measure is not necessarily equivalent to the

measurement of an unconscious construct. Participants may be unaware

what the task assesses, but that does not necessarily mean that they are

unaware of their attitudes or evaluations (Fazio & Olson, 2003). Thus, the

term automatic is not equivalent to the term unconscious, but rather means

that the employed indirect measures leave insufficient time for participants

to strategically control their response. To avoid confusing terminology, we

follow Fazio and Olson (2003) (see e.g. MacLeo, 1989) in using the term

indirect measure instead of implicit measure, and the term direct measure

instead of explicit measure, because ‘implicit’ seems to carry the notion of

‘unconscious’, while ‘explicit’ seems to carry the notion of ‘conscious’.

Whereas Lamote et al. (2004) showed that the

affective priming paradigm is useful for assessing food

likes and dislikes in general, the major aim of the current

experiments is to study whether this paradigm and a

related indirect measure are sensitive to individual

differences between restrained and unrestrained eaters in

food likes and dislikes. Heatherton, Herman, Polivy,

King, and McGree (1988, p. 19) define restrained eaters,

as selected by the Restraint Scale (Herman & Polivy,

1980), as dieters who ‘exhibit periods of restraint

punctuated by episodes of disinhibited overeating’.

Thus, restrained eaters have the intention of controlling

their weight, but often fail and indulge in high-fat

palatable foods that they normally do not allow

themselves to eat (Herman & Polivy, 1980, 2004). As

Gendall and Joyce (2001) suggest, the eating behavior of

restrained eaters might enhance the attractiveness of

these normally forbidden foods, because of psychological

frustration and temporary deprivation. Although evidence

is not unequivocal, several self-report studies (Gendall,

Joyce, Sullivan, & Bulik, 1998; Pelchat, 1997) found

more craving in restrained as compared to unrestrained

eaters. Moreover, some studies (Herman, Polivy, Klajner,

& Esses, 1981; Klajner, Herman, Polivy, & Chhabra,

1981; LeGoff & Spigelman, 1987) found evidence for

stronger physiological reactivity toward food cues in

restrained than unrestrained eaters, though again evidence

is not unequivocal (Nederkoorn & Jansen, 2002). The

current studies will investigate whether this possibly

greater craving and physiological reactivity in restrained

eaters is accompanied by a greater liking of these foods

on a relatively automatic level.

Prior research in our laboratory also focused on

individual differences in food likings, but used another

indirect measure (Roefs & Jansen, 2002), the Implicit

Association Test (IAT; Greenwald, McGhee, & Schwartz,

1998). Contrary to the hypothesis that an obese group

would have a stronger liking of high-fat foods than of

low-fat foods than a lean control group (see e.g.

Drewnowski, Brunzell, Sande, Iverius, & Greenwood,

1985; Rissanen et al., 2002), both groups showed more

negative associations with high-fat foods than with low-

fat foods, with this effect being more pronounced for the

obese group. The unexpected results of Roefs and Jansen

(2002) can probably most easily be explained by the

specific characteristics of the IAT, in that the fat content

of the food is very salient for participants in this task.

The results of De Houwer’s studies (2001, 2003b)

suggest that IAT effects are strongly influenced by the

basis of categorization (in our case fat content: high-fat

vs. low-fat). It is perhaps unsurprising that people do not

like the fact that their favorite food is high-fat, though

they obviously like the taste of it.

The advantage of both the priming task (Fazio et al.,

1986) and another, recently developed indirect measure,

the extrinsic affective Simon task (EAST; De Houwer,

A. Roefs et al. / Appetite 44 (2005) 103–114 105

2003a), is that these tasks do not demand that the food

be categorized in terms of fat content, thereby avoiding a

specific focus on the fat content (cf. IAT). Thus, indirect

measures may not all assess the same underlying

construct (Bosson, Swann, & Pennebaker, 2000; Fazio

& Olson, 2003; Olson & Fazio, 2003; but see Cunning-

ham, Preacher, & Banaji, 2001). Because of these and

other disadvantages of the IAT (De Houwer, 2002) we

chose to continue our research on individual differences

in food likings using other indirect measures. In

Experiment 1, the affective priming paradigm will be

employed (Fazio et al., 1986), following the lead of

Lamote et al. (2004). In Experiment 2, a recently

developed paradigm, the EAST (De Houwer, 2003a),

will be employed, to seek convergence and generalization

in a different paradigm. Similar to the affective priming

paradigm, participants are not required to categorize the

foods into pre-defined categories. Both experiments

investigate whether restrained eaters display a stronger

liking of palatable foods over unpalatable foods than

unrestrained eaters, and whether this is specific for high-

fat palatable foods, the foods that restrained eaters

normally do not allow themselves to eat.

Experiment 1

Method

Participants

The participants were female introductory psychology

students at the University of Toronto. They were selected

on the basis of their score on the Restraint Scale

(Herman & Polivy, 1980) from a large group of students

who had indicated that they would like to participate in

research, and they took part either for bonus credit in a

course or for $10. Thirty-two participants were classified

as restrained eaters, indicated by a score of 15 or higher

on the Restraint Scale (age: MZ19.5, SDZ2.0; self-

reported weight (kg): MZ63.0, SDZ10.2; BMI: MZ22.9, SDZ3.3, RangeZ17.1–30.8; total score Restraint

Scale: MZ20.0, SDZ3.6). Thirty-seven participants were

classified as unrestrained eaters, indicated by a score of

14 or lower on the Restraint Scale (age: MZ19.5, SDZ1.8; self-reported weight (kg): MZ55.7, SDZ7.7; BMI:

MZ20.6, SDZ2.6, RangeZ16.5–28.2; total score

Restraint Scale: MZ8.1, SDZ3.3). BMI refers to body

mass index, which is simply the ratio of weight to

squared height (kg/m2). The two groups did not differ

significantly in age, t(67)!1. However, they did differ in

BMI, t(67)Z3.24, p!0.01. The data of one additional

participant (restrained eater) were deleted from all

analyses because of a high percentage (OMC3SD) of

trials with errors or responses that were either too slow

(O2000 ms) or too fast (!200 ms).

Stimulus selection and timing of trials in the priming task

Stimuli. Sixteen high-fat food words (e.g. ‘chocolate’)

and 16 low-fat food words (e.g. ‘melon’) served as

primes (see Appendix A). The two groups of stimuli did

not differ significantly in word length (number of letters:

high-fat: MZ6.3, SDZ1.4 vs. low-fat: MZ6.1, SDZ1.5), t(30)!1. These foods were selected on the basis of

both a food table (Nevo Tabel, 1993) and prior research

related to food and restraint (McCabe, 1999).

Thirty-two general positive (e.g. ‘love’) and 32

general negative (e.g. ‘dreadful’) words served as targets

(see Appendix A), and were selected according to norms

by Bellezza, Greenwald, and Banaji (1986), who had

participants rate a large number of words on 5-point

scales for pleasantness (very unpleasant—very pleasant),

visual imagery (no image at all—very clear and vivid

visual image), and familiarity (encounter very infre-

quently—encounter very frequently). The two groups of

stimuli differed significantly in pleasantness (negative:

MZ1.4, SDZ0.1 vs. positive: MZ4.6, SDZ0.1, t(62)Z118.29, p!0.001. There were no significant differences

between the two groups of stimuli in visual imagery

(negative: MZ3.6, SDZ0.5 vs. positive: MZ3.7, SDZ0.5), t(62)!1, affective extremity (negative: MZ1.6,

SDZ0.1 vs. positive: MZ1.6, SDZ0.1), t(62)Z1.39,

pZ0.17, or word length (negative: MZ6.4, SDZ1.2 vs.

positive: MZ6.1, SDZ1.7), t(62)!1. However, there

was a significant difference in familiarity ratings

(negative: MZ2.8, SDZ0.6 vs. positive: MZ3.5, SDZ0.6), t(62)Z4.73, p!0.001. Primes and targets were

presented in black lower-case letters (6 mm high) against

a light background on a computer monitor.

Randomization of stimuli. Each of the 3 blocks

consisted of 64 trials, resulting in a total of 192 trials.

Each food-stimulus (prime) was paired once with a

positive and once with a negative target in each block.

All primes had been presented once (with either a

positive or a negative target) before any prime was

presented for the second time in each block. For each set

of two participants (one restrained and one unrestrained

eater), and for each of the three blocks, it was

determined randomly which half of the high-fat foods

and which half of the low-fat foods would be paired first

with a positive/negative target. Primes and targets were

both randomly selected (uniquely for each participant)

from their respective sets without replacement.

Trial timing. The timing of trials was modeled after

the procedure of Hermans, De Houwer, et al. (2001).

Each trial started with a warning tone (200 ms), followed

by a fixation cross (500 ms). Then, the prime was

presented for 200 ms. After a 150 ms stimulus onset

asynchrony (SOA)—the time that elapses between the

onset of the prime and the onset of the target—the target

was presented on the monitor. Thus, prime and target co-

occurred for 50 ms. An SOA of 150 ms was chosen

because Hermans, De Houwer et al. (2001) showed that

A. Roefs et al. / Appetite 44 (2005) 103–114106

this SOA was optimal for the expression of affective

priming effects. Because of the 50 ms simultaneous

presentation, the prime was presented 4.5 mm above

the center of the monitor, and the target 4.5 mm below

the center of the monitor (as measured to the center-line

of the characters).2 The target remained on the monitor

until a response was given or for 2500 ms if no response

was given. If an error was made or a response was either

too slow or too fast, or if no response at all was given, a

warning appeared on the screen for 300 ms. The inter-

trial interval was 2500 ms.

Direct measures

Direct rating task of foods. The 32 foods that were used

as primes were presented in random order (identical for all

participants) on a paper-and-pencil rating task. Participants

were asked to rate the foods on taste (5-point scale: 1:

dislike a lot – 5: like a lot).

Direct rank order task. Participants were given two

separate lists, one with the high-fat foods (16) and one with

the low-fat foods (16), presented in the same random order

for all participants. They were asked to rank each list of

foods in order of how much they liked the taste, from the

most tasty item to the least tasty item. The high-fat and low-

fat foods were presented on separate lists to minimize

influences of socially desirable answering tendencies. If

these two types of food had been presented mixed in one list,

participants might have been ‘tempted’ to give the low-fat

foods a better rank, even though they were instructed to pay

attention only to the taste of the foods.

This direct ranking task was included specifically for the

purpose of stimulus selection for the priming task. Because

the palatability of foods likely varies widely among

participants (Frank & van der Klaauw, 1994), we decided

to determine prime palatability uniquely for each partici-

pant. On the basis of the direct rank order task, we selected

the 5 most liked items, and the 5 least liked items for each

type of food (high-fat and low-fat) for each participant. To

prevent the direct ranking task from interfering with the

priming task, all primes (32) were initially included in the

priming task. In the analyses, we used only those trials on

which one of the idiosyncratically selected 20 primes

appeared (40 trials per block). Selecting primes in this way

provided us with 5 high-fat palatable foods, 5 low-fat

palatable foods, 5 high-fat unpalatable foods, and 5 low-fat

unpalatable foods per participant.

Restraint Scale. The Restraint Scale measures ‘the extent

to which participants showed evidence of dieting and

concern about their weight’ (Herman & Polivy, 1980,

2 Note however that the uncentered presentation mode might have

underestimated a possible priming effect involving fat content (Hermans,

De Houwer, et al., 2001). This way of presenting stimuli was chosen,

because originally this experiment included a second part in which the SOA

was varied, but in which the prime presentation was to be held constant at

200 ms.

p. 212). The maximum score on this scale is 35, whereas the

minimum score is 0.

Procedure

Participants were tested individually in a quiet

research room. For the affective priming task, they

were instructed to read the first word silently, and then to

decide whether the second word was positive or negative,

pressing the corresponding key on the key box (key

assignment was counterbalanced across participants).

They were told to respond as quickly as possible but

to avoid making too many mistakes. They were then

presented with 16 practice trials using stimulus materials

different from those on the experimental trials. Similarly

to Zack, Toneatto, and MacLeod (1999), they were given

a free recall test for the primes immediately after the

practice trials, writing down as many words as they

could remember that appeared as a first word (prime) on

the computer task that they had just performed. This task

was included to ensure that participants paid attention to

both primes and targets by raising the possibility that

participants might be asked to perform such a free recall

task later in the procedure. After this memory test, a

brief reminder appeared on the monitor concerning how

to perform the priming task. Participants were now ready

to begin the actual priming task, which was made up of

three blocks with a short break after each block.

After the priming task, the participants were again given

a free recall test for the primes. Subsequently, a manipu-

lation check was performed, to check whether participants

realized what the computer task assessed. Next, participants

were asked to perform the rating and rank-ordering of the

foods as described in Section 2.1.3. Their final task was to

complete the Restraint Scale.

Apparatus

The experiment was carried out on a Dell Inspiron 5000e

notebook computer with a Pentium III processor, connected

to a Samsung SyncMaster 750S monitor. Key responses

were registered by an external response device with better

than 1 ms accuracy. The software controlling the exper-

iment was programmed in ERTS (Experimental Run Time

System, Beringer, 1996).

Design and analysis

Data were analyzed using a 2(target affect: positive vs.

negative)!2(fat content prime: high-fat vs. low-fat)!2(palatability of prime: palatable vs. unpalatable)!2(group: restrained vs. unrestrained) analysis of variance

(ANOVA), with repeated measures on the first three factors.

An interaction between prime and target indicates that there

is a priming effect (i.e. that the presentation of the prime

influenced the speed and accuracy of responding to

the target). Before analyzing the data, we discarded

responses that were either too fast (!200 ms) or too

slow (O2000 ms), a total of only 0.19% of all trials.

Fig. 1. Experiment 1: The palatability priming effect as a function of

participant group (restrained vs. unrestrained eaters) and fat content of the

prime. The palatability priming effects for each participant and separately

for high-fat and low-fat foods are computed as: ((mean (palatableK)C

mean (unpalatableC))/2)K((mean (palatableC)Cmean (unpalatableK))/2). In this formula, ‘palatable’ or ‘unpalatable’ indicate the palatability of

the prime, whereas the plus or minus sign right behind it, indicate whether

the target was positive or negative. A positive score indicates a liking of

palatable foods over unpalatable foods, and a negative score indicates a

liking of unpalatable foods over palatable foods. Error bars represent one

standard error. Note: RSZrestrained eaters; URSZunrestrained eaters;

HFZhigh-fat food; LFZlow-fat food.

Table 1

Experiment 1: Mean palatability ratings (1: very unpalatable – 5: very

palatable) of the direct rating task

Food Mean SD

A. Roefs et al. / Appetite 44 (2005) 103–114 107

Response latencies associated with error responses (4.5%)

were also discarded. All reported analyses are for response

latencies as the dependent variable. Analyses on error

percentages will not be reported because they did not

produce relevant significant results.

Results

Affective priming task3

The speed of the response to the target was influenced by

the palatability of the prime in the expected direction:

participants were on average faster on congruent (palata-

ble—positive/unpalatable—negative) than on incongruent

(palatable—negative/unpalatable—positive) trials. This

prime palatability!target affect interaction was significant,

F(1, 67)Z5.87, p!0.05, and qualified main effects of both

target affect, F(1, 67)Z9.48, p!0.01, and prime palatability,

F(1, 67)Z5.11, p!0.05 (see Fig. 1). This interaction effect

(i.e. priming effect) means that participants displayed a liking

of palatable foods over unpalatable foods on a relatively

automatic level. However, this effect did not seem to be

different for the restrained and unrestrained groups, as

indicated by the non-significant target affect!prime palat-

ability!group interaction, F(1, 67)!1. Moreover, the fat

content did not seem to influence the results. None of the

relevant interactions—target affect!prime fat content (!group) and target affect!prime fat content!prime palat-

ability (!group)—was significant, all Fs(1, 67)!1.39. Nor

were the main effects of prime fat content, F(1, 67)!1, or

group F(1, 67)Z1.88, pZ0.18, significant.

Direct rating task palatability

Data for ratings of all 32 foods were analyzed. In a 2 (fat

content: high-fat vs. low-fat)!2(group: restrained vs.

unrestrained) ANOVA. On the direct task, participants in

general indicated that they preferred the tastes of low-fat

foods to those of high-fat foods, F(1, 67)Z11.46, p!0.01, a

finding which may partly reflect disapproval of high-fat

foods, rather than genuine dislike. Neither the main effect of

group, nor the fat content!group interaction was signifi-

cant, both Fs(1, 67)!1 (Table 1).

An analysis was also performed on just the five most

palatable high-fat and low-fat items and the five most

unpalatable high-fat and low-fat items, to examine whether

3 In the reported analyses, one participant was excluded because of a high

percentage (OMC3SD) of trials with errors or responses that were either

too slow (O2000 ms) or too fast (!200 ms). Analyses including this

participant did not affect the latency results. In the error analyses, however,

the target affect!prime palatability!group interaction was marginally

significant (pZ0.07). In separate analyses for the two groups, the target

affect!prime palatability interaction was not significant for the restrained

eaters (F!1), but was marginally significant for the unrestrained eaters

(pZ0.07). Unrestrained participants made fewer errors on incongruent

trials (palatable prime—negative target/unpalatable prime—positive target)

than on congruent trials (palatable prime—positive target/unpalatable

prime—negative target).

the high-fat and low-fat foods differed in palatability rating

within these extreme fives, in a 2(palatability: top five

palatable vs. top five unpalatable)!2(fat content: high-fat

vs. low-fat)!2(group) ANOVA. The palatability factor was

based on the responses on the rank order task. Participants

strongly preferred the palatable items over the unpalatable

items, F(1, 67)Z767.33, p!0.001, which is in accordance

with the findings on the affective priming task. Participants

again also preferred the low-fat foods over the high-fat

RS URS RS URS

HF 3.8 3.8 0.6 0.5

LF 4.1 4.0 0.4 0.5

HFC 4.7 4.7 0.4 0.3

HFK 2.6 2.6 0.9 0.8

LFC 4.8 4.8 0.2 0.3

LFK 3.0 3.0 0.7 0.9

Standard deviations (SD) are reported in a separate column. The high-fat

and low-fat foods are grouped into palatable and unpalatable foods based on

the direct rank order task. Note: HFZ all HF foods; LFZ all low-fat foods;

HFCZhigh-fat palatable foods; HFKZ high-fat unpalatable foods;

LFCZ low-fat palatable foods; LFKZ low-fat unpalatable foods; RSZrestrained eaters, URSZ unrestrained eaters.

A. Roefs et al. / Appetite 44 (2005) 103–114108

foods, F(1, 67)Z14.18, p!0.001. Moreover, the palat-

ability!fat content interaction was significant, F(1, 67)Z7.92, p!0.01. Further analyses indicate that the effect of fat

content was larger for unpalatable items, F(1, 67)Z13.47,

p!0.001, than for palatable items, F(1, 67)Z4.41, p!0.05

(see Table 1).

Experiment 2

We sought to corroborate the findings of Experiment 1 in

a second experiment, which tested the same hypothesis in a

different, converging way. Experiment 2 differed in three

important ways from Experiment 1. First, to study whether

the findings of Experiment 1 might have been method

specific, a different indirect measure, the EAST (De

Houwer, 2003a), was used in Experiment 2. Comparable

to the affective priming paradigm, this paradigm also does

not demand that the food stimuli be categorized in pre-

defined categories (high-fat vs. low-fat). Second, in

Experiment 2, the palatability factor was determined not

only on an individual basis, as in Experiment 1, but also on

the basis of a separate pilot study, conducted with a different

group of participants. So, using the pilot data, the four

groups of stimuli were determined on an a priori basis. This

manipulation permits comparison of these two methods of

stimulus selection. The advantage of an idiosyncratic

measure of palatability for stimulus selection is that this

procedure guarantees that the selected stimuli fit with the

participant. The advantage of stimulus selection based on a

pilot study is that the stimuli that are used in the analyses are

the same for all participants. Finally, a third modification

involved how the positive and negative stimuli (in this

experiment the white words) were selected. In Experiment

1, the target words were simply general positive and

negative words. These sets of stimuli were chosen to avoid

biasing the participants, and thus their responses to the

primes (foods). In Experiment 2, synonyms of the concepts

‘palatable’ and ‘unpalatable’ were used as the white stimuli,

in an attempt to make the (un)palatability a salient feature.

Method

Pilot study palatability

The goal of this pilot study was to determine which foods

are generally liked and disliked by female university

students (nZ64). Participants were given a list of 28 high-

fat foods and a list of 33 low-fat foods, and were asked to

consider each list separately. If high-fat and low-fat foods

had been mixed, participants might have been tempted to

select mainly low-fat foods, because a liking for low-fat

foods may be seen as socially desirable. They were asked to

choose from each list the eight foods that they liked most

and the eight foods that they disliked most. Then, they

were asked to rank the eight foods they (dis)liked most

from 1 ((dis)like the most) to 8 ((dis)like the least).

Stimulus selection was based on the mean ranking of a

food, weighted by the number of participants who put that

food in their selection of (dis)liked foods. In this way, we

selected 5 high-fat liked, 5 high-fat disliked, 5 low-fat liked,

and 5 low-fat disliked foods (see Appendix A).

Participants

The participants were female students at Maastricht

University in the Netherlands. Participants were selected in

the same way as in Experiment 1. All participants were paid

V 7.50. Twenty-six participants, scoring 15 or higher on the

Restraint Scale, were classified as restrained eaters (age:

MZ19.6, SDZ2.2; weight (kg): MZ69.0, SDZ13.2; BMI:

MZ24.6, SDZ4.16, RangeZ19.8–34.4; total score

Restraint Scale: MZ20.3, SDZ4.0). Thirty participants,

scoring 14 or lower on the Restraint Scale, were classified as

unrestrained eaters (age: MZ19.3, SDZ1.1; weight (kg):

MZ63.6, SDZ7.8; BMI: MZ22.3, SDZ2.4, RangeZ18.6–27.1; total score Restraint Scale: MZ10.1, SDZ2.9).

The experimenter measured body weight and height at the

end of the experiment. Two restrained eaters did not want to

be weighed, and did not want to report their weight either.

The two groups did not differ significantly in age, t(54)!1.

They did differ significantly in BMI, t(34.95)Z2.43,

p!0.05.

Overview extrinsic affective Simon task (EAST)

In the critical blocks of the EAST (De Houwer, 2003a),

white and colored (blue vs. green) stimuli were presented

intermixed. Participants had to categorize the white words

based on the meaning of these words, and the colored words

based on their color. The white stimuli were synonyms of

the concept palatable or the concept unpalatable. By

assigning one type of white stimuli (e.g. synonyms of

palatable) to the left response key, and the other type of

white stimuli (e.g. synonyms of unpalatable) to the right

response key, participants extrinsically associated each key

with a specific type of stimuli. Key-assignment was

counterbalanced over participants.

The colored stimuli were the concepts of interest, and

consisted of the same four types of food as in Experiment 1.

Each food stimulus was presented once in blue and once in

green in each experimental block. The participant was

instructed to respond to one color (e.g. blue) with the left

response key, and to respond to the other color (e.g. green)

with the right response key. Again, key assignment was

counterbalanced over participants. In this example, it should

be easier for a participant to categorize the color of a

palatable food when it is presented in blue rather than green,

because both blue food words and white synonyms of the

concept palatable map onto the same response key (left).

Correspondingly, it should be easier for a participant to

categorize the color of an unpalatable food when it is

presented in green rather than blue, because both green food

words and white synonyms of the concept unpalatable map

onto the same response key (right). The dependent variables

A. Roefs et al. / Appetite 44 (2005) 103–114 109

in this task were the response latency and the percentage of

errors. The EAST effect is defined as the difference in

response latency and/or error percentages when responding

with the extrinsically incongruent key versus the extrinsi-

cally congruent key.

Stimulus selection and timing of the trials in the EAST

Stimuli. Five high-fat palatable (HFC), five high-fat

unpalatable (HFK), five low-fat palatable (LFC), and five

low-fat unpalatable (LFK) foods were selected as colored

stimuli, on the basis of the pilot task (see Appendix A).

These four groups of stimuli did not differ significantly in

word length (number of letters: HFC: MZ6.2, SDZ1.8;

HFK: MZ6.2, SDZ1.9; LFC: MZ6.4, SDZ2.2; LFK:

MZ7.2, SDZ1.1), F(3, 16)!1.

The white stimuli were 5 synonyms for the concept

palatable and 5 synonyms for the concept unpalatable (see

Appendix A). These stimuli were selected using the online

version of the Van Dale Dictionary for the Dutch language.

The two groups of stimuli did not differ significantly in word

length (number of letters: synonyms palatable: MZ8.0,

SDZ2.7; synonyms unpalatable: MZ8.4, SDZ3.4), t(8)!1. Stimuli were presented against a black background.

Following De Houwer (2003a), using the RGB color

system, the blue color was defined as 0% red, 38% green,

and 46% blue, whereas the green color was defined as 0%

red, 46% green, and 38% blue. The resulting blue and green

colors were distinguishable but rather similar, as was

intended.

Randomization of stimuli. In the first block, the 10 white

words were presented twice in a random order unique for

each participant. Stimuli were drawn without replacement,

and the set was initialized when it was empty. In the second

block, each food word was presented once in a random order

unique for each participant. Half of the foods were presented

in green, the other half in blue. The color (blue or green) of

the presented stimulus was also determined randomly and

was unique for each participant.

The four experimental blocks followed, in each of which

there were 20 white trials and 40 colored trials. The colored

trials were the crucial trials, and all food words were

presented once in each color, in each block. All words were

presented once in either green or blue before any words

were presented for the second time in the other color. For

each pair of two participants (one restrained and one

unrestrained eater), and for each of four experimental

blocks, it was determined randomly which half of the

stimuli would be presented first in green, and which half

would be presented first in blue. Again, all stimuli were

drawn randomly from their respective sets without replace-

ment, unique for each participant and each block. Sets were

initialized when they were empty. Following De Houwer

(2003a), each test block started with a few warm-up trials

(four in the first test block, two in the other three test

blocks). Each of the 10 white words appeared on one of

these warm-up trials. The selection of the white words for

the warm-up trials was random (unique for each partici-

pant). In each block, half of the warm-up trials were positive

and the other half were negative.

Trial timing. The timing of the trials was modeled after

that of De Houwer (2003a). Each trial started with a white

fixation cross in the middle of the monitor (500 ms). Then,

the stimulus was presented in the color appropriate for that

trial (white, green, or blue), also in the middle of the

monitor. The stimulus remained on the monitor until the

participant responded, or until 10 s had elapsed. The inter-

trial interval was 1500 ms. If an error was made, or a

response was either too slow (O3000 ms) or too fast (!300 ms), or if no response at all was given, a warning

appeared on the monitor for 300 ms. We changed the cut-

offs in this experiment relative to Experiment 1, to closely

follow De Houwer’s (2003a) procedure, thereby also

generalizing presentation parameters.

Direct measures

The same kind of direct measures were used as in

Experiment 1. Of course, this time, only the 20 foods that

appeared in the EAST were presented on these direct

measures. In the paper-and-pencil rating task, participants

were asked to rate the foods on taste (7-point scale: dislike a

lot–like a lot).

The ranking task was the same as in Experiment 1 and

was used as an idiosyncratic measure of palatability.

EAST data were then analyzed in two ways: with a

palatability factor based on the pilot study (all 5 stimuli

for each of 4 food types), and with a palatability factor

based on this idiosyncratic measure. For the idiosyncratic

measure of palatability, the three most palatable and the

three most unpalatable high-fat and low-fat foods were

selected for each participant and used in the analyses. A

final direct measure was the Restraint Scale (Herman &

Polivy, 1980).

Procedure

Participants were tested individually in a quiet, dimly lit

research room. First, general instructions were given about

which key to press to what kind of stimulus in the EAST.

Responses had to be given as quickly as possible, avoiding

too many mistakes. Before each block, the relevant

instructions were repeated on the monitor. After each

block, participants could take a quick break. To control for

time of day effects (e.g. time elapsed since lunch or

breakfast), average start time of the test session did not

differ significantly between the two groups, t(54)!1. After

the EAST, participants were asked to perform the rating and

rank order task on the foods. Then, they were asked to fill

out the Restraint Scale. Afterward, participants were briefly

asked about their dieting history, eating habits, and medical

condition. Finally, the experimenter measured the partici-

pant’s height and weight.

A. Roefs et al. / Appetite 44 (2005) 103–114110

Apparatus

The experiment was carried out on a Dell Optiplex GX

110 computer with a Pentium III processor, connected to a

Dell monitor. The same external response device and the

same software package as in Experiment 1 were used.

Design and analysis

EAST results were analyzed in 2(fat content food words:

high-fat vs. low-fat)!2(palatability food words: palatable

vs. unpalatable)!2(extrinsic response: synonym palatable

vs. synonym unpalatable)!2(group: restrained vs. unrest-

rained eaters) ANOVAs, with repeated measures on the first

three factors. An interaction between a food factor and the

extrinsic response indicates a significant EAST effect. All

analyses were performed on the colored trials only, the trials

on which food words were presented. Using the same cut-off

values as De Houwer (2003a) did for his EAST studies,

responses that were either too fast (!300 ms) or too slow

(O3000 ms) were discarded, a total of only 0.28% of all

trials in the analysis with the pilot based palatability factor,

and 0.17% of all trials for the analysis with the idiosyncratic

palatability factor. Response latencies associated with error

responses (pilot-based: 4.7%; idiosyncratic measure: 4.3%)

were also discarded.

Fig. 2. Experiment 2: The EAST effect as a function of participant group (restrained

The EAST effect (RT and percentage of errors) for each participant and separate

extrinsic palatable response). A positive score indicates a relative liking of the type

food. Error bars represent one standard error. Panel A represents the results for

represents the results for the analyses with the palatability factor based on the id

latencies, whereas the right panels represent the results for the error percentages.

palatable food; LFCZ low-fat palatable food; HFKZ high-fat unpalatable food

Results and discussion

Palatability based on pilot study

Response latencies. No significant effects emerged from

this analysis. The EAST effect, palatability!extrinsic

response, F(1, 54)Z2.40, pZ0.13, was not significant

(see Fig. 2A).

Percentage errors. Unrestrained eaters made fewer

errors as compared to restrained eaters, F(1, 54)Z4.38,

p!0.05. Moreover, the EAST effect, palatability!extrinsic

response was significant and in the expected direction, F(1,

54)Z8.81, p!0.01. On colored trials with palatable items,

participants made fewer errors when responding with the

extrinsic palatable response than when responding with the

extrinsic unpalatable response, F(1, 54)Z4.92, p!0.05,

whereas the opposite was true for colored trials with

unpalatable items. Participants then made fewer errors when

responding with the extrinsic unpalatable response, F(1,

54)Z5.54, p!0.05. However, the EAST effect (palatabil-

ity!extrinsic response) was not modified by the fat content

of the foods, F(1, 54)!1, the restraint-status of the

participant, F(1, 54)!1, or an interaction between

restraint-status and fat content, F(1, 54)!1 (see Fig. 2A).

vs. unrestrained eaters), palatability of the food, and fat content of the food.

ly for each type of food is computed as: (extrinsic unpalatable responseKof food, whereas a negative score indicates a relative disliking of the type of

the analyses with the palatability factor based on the pilot study. Panel B

iosyncratic rankings. The left panels represent the results for the response

Note: RSZ restrained eaters; URSZ unrestrained eaters; HFCZ high-fat

; LFKZ low-fat unpalatable food.

A. Roefs et al. / Appetite 44 (2005) 103–114 111

Idiosyncratic measure of palatability

Response latencies. A fat content!palatability!extrin-

sic response interaction, F(1, 54)Z6.06, p!0.05, qualified

a palatability!extrinsic response interaction, F(1, 54)Z7.77, p!0.01. For high-fat foods, the palatability!extrinsic

response interaction was not significant, F(1, 54)!1. For

low-fat foods, the palatability!extrinsic response inter-

action was significant, F(1, 54)Z11.78, p!0.01. On

colored trials with unpalatable items, participants were

faster when responding with the extrinsic unpalatable

response than when responding with the extrinsic palatable

response, F(1, 54)Z8.57, p!0.01. For palatable items, this

main effect of extrinsic response was not significant, F(1,

54)Z1.63, pZ0.21 (see Fig. 2B). This three-way inter-

action was not modified by restraint-status, F(1, 54)!1.

Percentage errors. The EAST effect, palatability!extrinsic response, was significant, F(1, 54)Z12.71, p!0.01. On colored trials with palatable items, participants

made fewer errors when responding with the extrinsic

palatable response than when responding with the extrinsic

unpalatable response, F(1, 54)Z5.97, p!0.05, whereas the

opposite was true for colored trials with unpalatable items.

Participants then made fewer errors when responding with

the extrinsic unpalatable response, F(1, 54)Z9.61, p!0.01.

However, the EAST effect (palatability!extrinsic

response) was not modified by the fat content of the

foods, F(1, 54)!1.14, the restraint-status, F(1, 54)!1 or an

interaction between fat content and restraint-status, F(1,

54)!1 (see Fig. 2B).

Direct rating task palatability

Data were analyzed in a 2(fat content: high-fat vs. low-

fat)!2(palatability: palatable vs. unpalatable)!2(group:

restrained vs. unrestrained eaters) ANOVA. The palatability

factor was based on the pilot study. Unsurprisingly, the

palatable items were judged to be more palatable than the

unpalatable items, F(1, 54)Z318.10, p!0.001, which is in

accordance with the findings on the EAST. A fat content!palatability interaction, F(1, 54)Z9.53, p!0.01, qualifying

the main effect of palatability, suggests that the difference

between palatable and unpalatable items was larger for the

high-fat foods than for the low-fat foods (see Table 2).

Table 2

Experiment 2: Mean palatability ratings (1: very unpalatable – 7: very

palatable) of the direct rating task

Food Mean SD

RS URS RS URS

HFC 6.0 6.1 0.8 0.6

HFK 3.6 3.9 1.0 1.0

LFC 5.9 5.9 0.5 0.6

LFK 4.2 4.1 1.0 1.0

Standard deviations (SD) are reported in a separate column. Note: HFCZhigh-fat palatable foods; HFKZ high-fat unpalatable foods; LFCZ low-

fat palatable foods; LFKZ low-fat unpalatable foods; RSZ restrained

eaters, URSZ unrestrained eaters.

Again, results were not influenced by the restraint-status of

the participants, Fs(1, 54)!1.40.

General discussion

The goal of the current experiments was to test the

hypothesis that restrained eaters would be characterized by a

stronger liking of (high-fat) palatable foods. To do so, we

employed two indirect measures: the affective priming

paradigm (Fazio et al., 1986) and the EAST (De Houwer,

2003a), seeking to provide convergence and generalization.

In Experiment 1, the affective priming task, participants

were faster on congruent (palatable—positive/unpalata-

ble—negative) trials than on incongruent (palatable—

negative/unpalatable—positive) trials. In Experiment 2

(EAST), participants responded more accurately to pala-

table food words when these food words—which had to be

categorized based on their color (green vs. blue)—were

mapped onto the same response key as synonyms of the

concept palatable. Similarly, participants responded more

accurately to unpalatable food words when these food

words—which had to be categorized based on their color—

were mapped onto the same response key as synonyms of

the concept unpalatable. Moreover, in Experiment 2, results

were not substantially affected by the way that the palatable

and unpalatable stimuli had been selected (idiosyncratically

vs. pilot study). EAST effects were a little more consistent

for the low-fat foods, in that in the RT analyses (idiosyn-

cratic measure of palatability) an EAST effect was only

observed for the low-fat foods. In the analyses on error

percentages, EAST effects were not influenced by the fat

content of the foods.

The results of both experiments suggest that people in

general can and do evaluate the palatability of foods quite

automatically. The evaluation of the palatability of the food

is automatic in the sense that the two paradigms did not

allow much time for controlled processing because of the

relatively short presentation durations and quick respond-

ing. The hypothesis that restrained eaters would show a

stronger liking for (high-fat) palatable foods was not

supported in either experiment. Note, however, that the

conclusion of no differences between restrained and

unrestrained eaters hinges on a null-finding, so power may

have been a problem. But it is worth noting that the finding

that people make this early response toward food based on

its taste is consistent with prior research. Hermans, Baeyens,

and Lamote (2001) and Lamote et al. (2004) found an

affective priming effect for foods based on general liking for

these foods.

Why did we not find evidence for the principal

hypothesis that restrained eaters would show a stronger

liking for (high-fat) palatable foods? If it is assumed that

greater craving necessarily is accompanied by greater liking

(for a discussion of this assumption, see Berridge, 1996),

then the results of the current experiments could be taken as

evidence that restrained eaters are not characterized by

A. Roefs et al. / Appetite 44 (2005) 103–114112

differential craving responses toward high-fat palatable

foods. It could also be that the stimuli—names of foods—

that were used in the current studies were not strong enough

cues to elicit more craving in restrained eaters as compared

to unrestrained eaters, and thus did not lead to differential

liking responses.

An alternative explanation can be found in Berridge’s

(1996) theory of food reward. Though it is often assumed

that a craving for a food reflects liking the food, he

convincingly argues that this in fact might not always be the

case, and that there is no independent evidence for it. He

distinguishes between appetite and palatability, a distinction

between ‘the disposition to eat and the sensory pleasure of

actually eating’ (p. 4). Berridge prefers to use the common

word ‘wanting’ for the motivational effects of appetite and

the word ‘liking’ for the palatability effects. In Berridge’s

(1996) analysis, people can experience an increase in

wanting (craving), without experiencing an increase in

liking (increased palatability). Thus, our results could mean

that a possibly increased craving response toward high-fat

palatable foods in restrained eaters simply does not go

together with a stronger liking for the food, even if this

liking is assessed by an indirect measure. These indirect

measures may thus tap the palatability or liking of the food

independent of craving for that food.

Consequently, it might thus be the case that all people,

independent of restraint-status, like certain foods to the

same extent, but that restrained eaters have a stronger

wanting for those foods. An indication that restrained eaters

have a stronger wanting of certain foods is that—apart from

the studies cited in Section 1 suggesting that restrained

eaters have more cravings and stronger physiological

reactivity to food cues—they often do have a higher weight

and a higher energy intake (French et al., 1994; Jansen,

1996; Stice, 2002). Moreover, as Nederkoorn, van Eijs, and

Jansen (2004) recently showed, restrained eaters are more

impulsive than unrestrained eaters on a computer task that is

unrelated to eating behavior, the stop-signal task (Logan,

1994). As Nederkoorn et al. (2004) suggest, this impulsivity

might very well underlie their dysfunctional eating

behavior. In other words, restrained eaters might be less

able to resist the temptation of palatable food.

Wiers, van Woerden, Smulders, and de Jong (2002) draw

similar conclusions from their findings, based on the

incentive-sensitization theory (Robinson & Berridge,

2001). This theory resembles Berridge’s (1996) theory of

food reward. According to Robinson and Berridge (2001),

an addiction is driven by a stronger wanting (sensitized

arousal) of the drug or alcohol, and not by a greater liking of

it. Wiers et al. (2002) used an IAT (Greenwald et al., 1998),

and found that heavy drinkers associated alcohol more

strongly with arousal than did light drinkers. Heavy and

light drinkers did not differ in their valence associations:

they invariably associated alcohol with negative affect.

‘These arousal associations could reflect an appetitive

response directed toward the drug’ (alcohol) (Wiers et al.,

2002, p. 656).

Notably, in the current experiments, we found evidence

for positive associations with palatable foods. This was not

specific for restrained eaters or for high-fat foods, but—in

contrast to Wiers et al. (2002) study in which negative

alcohol associations were found—we found liking of

palatable foods over unpalatable foods. Importantly, Wiers

et al. (2002) used a different indirect measure, the IAT

(Greenwald et al., 1998). As explained in Section 1, these

indirect measures often do not correlate strongly (Bosson et

al., 2000; Fazio & Olson, 2003; Olson & Fazio, 2003),

which might suggest that these paradigms measure different

underlying constructs, although reliability and validity are

also concerns (Buchner & Wippich, 2000). Thus, the

difference in paradigms (IAT vs. affective priming and

EAST) might explain why Wiers et al. (2002)—like Roefs

and Jansen (2002)—found negative associations with the

supposedly craved substance, whereas the current exper-

iments found positive associations.

In sum, palatability, not fat content, determined respond-

ing for all individuals, regardless of their restraint-status. No

evidence was found for the hypothesis that restrained eaters

would show a greater liking of (high-fat) palatable foods. If

it is assumed that liking a food and craving for a food are

necessarily related, these findings could be taken as

evidence that restrained eaters are not characterized by

stronger craving responses specifically toward high-fat

palatable foods, as compared to unrestrained eaters.

However, that assumption appears doubtful. The wantin-

g/liking distinction of Berridge (1996) was suggested as a

relevant dimension that might help to explain the behavior

observed here, in that this theory states that craving and

liking might be independent processes. It thus may be that

restrained and unrestrained eaters do differ in craving

responses toward high-fat palatable foods, despite liking

these foods to the same extent. The employed indirect

measures might only be sensitive to liking responses

(palatability) and not to potential differences in craving.

At any rate, the factors involved in food likes and cravings,

and how they interact with the characteristics of the

individual eater, clearly represent a complex structure.

Acknowledgements

This study was financed by the Netherlands Research

Organization (NWO) Grant 425-20-801, awarded to Anita

Jansen. The authors would like to thank Reinout Wiers, Jorg

Huijding, Peter de Jong, and Martin Zack for advice on the

design of the current studies, Gerard van Breukelen for his

advice on the statistical analyses, Beatrijs Hauer for testing

the participants of Experiment 2, and Marieke Werrij and

Pascal van Gerven for their comments on an earlier draft of

this article. Correspondence concerning this article should

A. Roefs et al. / Appetite 44 (2005) 103–114 113

be addressed to Anne Roefs: a.roefs@psychology.uni-

maas.nl [A. Roefs].

Experiment 1

Positive targets: love, romantic, paradise, pleasure, joy,

laughter, cheer, humor, passion, terrific, enjoyment, happy,

caress, cuddle, honest, life, freedom, hug, peace, liberty,

treasure, triumph, loyal, sweetheart, truth, warmth, cozy,

glory, reward, flower, talent, honor.

Negative targets: killer, torture, devil, rape, brutal,

funeral, murderer, bastard, morgue, poison, burial, dreadful,

suicide, agony, failure, hatred, poverty, terrible, abuse,

unhappy, accident, hostage, despise, disaster, bankrupt, jail,

filth, slime, violent, grief, waste, tragedy.

High-fat foods: bacon, cake, cheese, chips, chocolate,

coconut, cookie, donut, fries, fudge, hamburger, herring,

nachos, pancake, peanuts, walnuts.

Low-fat foods: apple, banana, bread, broccoli, cabbage,

fruit, juice, melon, radish, popcorn, rice, spinach, straw-

berry, tomato, turkey, yogurt.

Experiment 2

Synonyms palatable: smakelijk (palatable), heerlijk

(delicious), lekker (tasty), verrukkelijk (delectable), zalig

(yummy).

Synonyms unpalatable: vies (vile), smerig (nasty),

afschuwelijk (horrible), walgelijk (disgusting), onsmakelijk

(unpalatable).

High-fat palatable foods: chocola (chocolate), chips

(chips), friet (fries), croissant (croissant), pizza (pizza).

High-fat unpalatable foods: haring (herring), speklap

(slice of bacon), pate (pate), boter (butter), pindakaas

(peanutbutter).

Low-fat palatable foods: aardbeien (strawberries), drui-

ven (grapes), meloen (melon), kip (chicken), popcorn

(popcorn).

Low-fat unpalatable foods: spruiten (Brussels sprouts),

witlof (chicory), zuurkool (sauerkraut), andijvie (endive),

radijs (radish).

References

Bargh, J. A., Chaiken, S., Govender, R., & Pratto, F. (1992). The generality

of the automatic attitude activation effect. Journal of Personality and

Social Psychology, 62, 893–912.

Bellezza, F. S., Greenwald, A. G., & Banaji, M. R. (1986). Words high and

low in pleasantness as rated by male and female college students.

Behavior Research Methods, Instruments, and Computers, 18, 299–

303.

Beringer, J. (1996). Experimental run time system (ERTS). [Computer

software]. Frankfurt, Germany: Berisoft.

Berridge, K. C. (1996). Food reward: Brain substrates of wanting and

liking. Neuroscience and Biobehavioral Reviews, 20, 1–25.

Bosson, J. K., Swann, W. B., Jr.., & Pennebaker, J. W. (2000). Stalking

the perfect measure of implicit self-esteem: The blind men and the

elephant revisited? Journal of Personality and Social Psychology,

79, 631–643.

Buchner, A., & Wippich, W. (2000). On the reliability of implicit and

explicit memory measures. Cognitive Psychology, 40, 227–259.

Cunningham, W. A., Preacher, K. J., & Banaji, M. B. (2001). Implicit

attitude measures: Consistency, stability, and convergent validity.

Psychological Science, 12, 163–170.

De Houwer, J. (2001). A structural and process analysis of the Implicit

Association Test. Journal of Experimental Social Psychology, 37, 443–

451.

De Houwer, J. (2002). The Implicit Association Test as a tool for studying

dysfunctional associations in psychopathology: Strengths and limi-

tations. Journal of Behavior Therapy and Experimental Psychiatry, 33,

115–133.

De Houwer, J. (2003a). The extrinsic affective Simon task. Experimental

Psychology, 50, 77–85.

De Houwer, J. (2003b). A structural analysis of indirect measures of

attitudes. In J. Musch, & K. C. Klauer (Eds.), The psychology of

evaluation: Affective processes in cognition and emotion (pp. 219–244).

Mahwah, NJ: Erlbaum.

Dovidio, J. F., Kawakami, K., & Gaertner, S. L. (2002). Implicit and

explicit prejudice and interracial interaction. Journal of Personality and

Social Psychology, 82, 62–68.

Drewnowski, A., Brunzell, J. D., Sande, K., Iverius, P. H., & Greenwood,

M. R. C. (1985). Sweet tooth reconsidered: Taste responsiveness in

human obesity. Physiology and Behavior, 35, 617–622.

Eertmans, A., Baeyens, F., & Van den Bergh, O. (2001). Food likes and

their relative importance in human eating behavior: Review and

preliminary suggestions for health promotion. Health Education

Research, 16, 443–456.

Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995).

Variability in automatic activation as an unobtrusive measure of racial

attitudes: A bona fide pipeline? Journal of Personality and Social

Psychology, 69, 1013–1027.

Fazio, R. H., & Olson, M. A. (2003). Implicit measures in social cognition

research: Their meaning and use. Annual Review of Psychology, 54,

297–327.

Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986).

On the automatic activation of attitudes. Journal of Personality and

Social Psychology, 50, 229–238.

Frank, R. A., & van der Klaauw, N. J. (1994). The contribution of

chemosensory factors to individual differences in reported food

preferences. Appetite, 22, 101–123.

French, S. A., Jeffery, R. W., Forster, J. L., McGovern, P. G., Kelder, S. H.,

& Baxter, J. E. (1994). Predictors of weight change over two years

among a population of working adults: The healthy worker project.

International Journal of Obesity, 18, 145–154.

Gendall, K. A., & Joyce, P. R. (2001). Characteristics of food cravers who

binge eat. In M. M. Hetherington (Ed.), Food cravings and addiction

(pp. 567–583). London: Leatherhead Publishing.

Gendall, K. A., Joyce, P. R., Sullivan, P. F., & Bulik, C. M. (1998). Food

cravers: Characteristics of those who binge. International Journal of

Eating Disorders, 23, 353–360.

Gerding, A. L., & Weinstein, L. (1992). Taste ratings of obese people, and

taste preferences based on geographical location. Bulletin of the

Psychonomic Society, 30, 509–510.

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998).

Measuring individual differences in implicit cognition: The Implicit

Association Test. Journal of Personality and Social Psychology, 74,

115–139.

Heatherton, T. F., Herman, C. P., Polivy, J., King, G. A., & McGree, S. T.

(1988). The (mis)measurement of restraint: An analysis of conceptual

and psychometric issues. Journal of Abnormal Psychology, 97, 19–28.

Herman, C. P., & Polivy, J. (1980). Restrained eating. In A. J. Stunkard

(Ed.), Obesity (pp. 208–225). Philadelphia, PA: Saunders.

A. Roefs et al. / Appetite 44 (2005) 103–114114

Herman, C. P., & Polivy, J. (2004). The self-regulation of eating:

Theoretical and practical problems. In R. F. Baumeister, &

K. D. Vohs (Eds.), Handbook of self-regulation research (pp. 492–508).

New York: Guilford Press.

Herman, C. P., Polivy, J., Klajner, F., & Esses, V. M. (1981). Salivation in

dieters and non-dieters. Appetite, 2, 356–361.

Hermans, D., Baeyens, F., & Lamote, S. (2001). Affective priming as an

implicit and unobtrusive measure of attitudes towards food items.

Unpublished research report, University of Leuven, Belgium.

Hermans, D., De Houwer, J., & Eelen, P. (1994). The affective priming

effect: Automatic evaluative information in memory. Cognition and

Emotion, 8, 515–533.

Hermans, D., De Houwer, J., & Eelen, P. (2001). A time course analysis of

the affective priming effect. Cognition and Emotion, 15, 143–165.

Jansen, A. (1996). How restrained eaters perceive the amount they eat.

British Journal of Clinical Psychology, 35, 381–392.

Klajner, F., Herman, C. P., Polivy, J., & Chhabra, R. (1981). Human

obesity, dieting, and anticipatory salivation to food. Physiology and

Behavior, 27, 195–198.

Klauer, K. C., & Musch, J. (2003). Affective priming: Findings and

theories. In J. Musch, & K. C. Klauer (Eds.), The psychology of

evaluation: Affective processes in cognition and emotion (pp. 7–49).

Mahwah, NJ: Erlbaum.

Lamote, S., Hermans, D., Baeyens, F., & Eelen, P. (2004). An exploration

of affective priming as an indirect measure of food attitudes. Appetite,

42, 279–286.

LeGoff, D. B., & Spigelman, M. N. (1987). Salivary responses to olfactory

food stimuli as a function of dietary restraint and body weight. Appetite,

8, 29–35.

Logan, G. D. (1994). On the ability to inhibit thought and action: A users’

guide to the stop signal paradigm. In D. Dagenbach, & T. H. Carr (Eds.),

Inhibitory processes in attention, memory, and language (pp. 189–239).

San Diego: Academic Press.

MacLeod, C. M. (1989). Directed forgetting affects both direct and indirect

tests of memory. Journal of Experimental Psychology: Learning,

Memory, and Cognition, 15, 13–21.

McCabe, R. E. (1999). Cognitive aspects of restraint: Situational and task

influences on memory for body shape-related information in dieters and

non-dieters. Unpublished doctoral dissertation, University of Toronto.

Mela, D. J. (1999). Symposium on ‘Functionality of nutrients and

behaviour’. Food choice and intake: The human factor. Proceedings

of the Nutrition Society, 58, 513–522.

Mela, D. J. (2001). Determinants of food choice: Relationships with obesity

and weight control. Paper presented at Food patterns in weight

management and health, Dallas, TX.

Nederkoorn, C., & Jansen, A. (2002). Cue reactivity and regulation of food

intake. Eating Behaviors, 3, 61–72.

Nederkoorn, C., Van Eijs, Y., & Jansen, A. (2004). Restrained eaters

act on impulse. Personality and Individual Differences

doi:10.1016/j.paid.2004.02.020.

Neely, J. H. (1991). Semantic priming effects in visual word recognition: A

selective review of current findings and theories. In D. Besner, &

G. W. Humphreys (Eds.), Basic processes in reading: Visual word

recognition (pp. 264–337). Hillsdale, NJ: Erlbaum.

Nevo Tabel, Nederlands voedingsstoffenbestand (1993). Den Haag, The

Netherlands [Dutch Food Composition Table].

Olson, M. A., & Fazio, R. H. (2003). Relations between implicit measures

of prejudice: What are we measuring? Psychological Science, 14,

636–639.

Palfai, T. P., & Ostafin, B. D. (2003). The influence of alcohol on the

activation of outcome expectancies: The role of evaluative expectancy

activation in drinking behavior. Journal of Studies on Alcohol, 64,

111–119.

Pelchat, M. L. (1997). Food cravings in young and elderly adults. Appetite,

28, 103–113.

Puhl, R., & Brownell, K. D. (2003). Ways of coping with obesity stigma:

Review and conceptual analysis. Eating Behaviors, 4, 53–78.

Rissanen, A., Hakala, P., Lissner, L., Mattlar, C. E., Koskenvuo, M., &

Ronnemaa, T. (2002). Acquired preference especially for dietary fat

and obesity: A study of weight-discordant monozygotic twin pairs.

International Journal of Obesity, 26, 973–977.

Robinson, T. E., & Berridge, K. C. (2001). Mechanisms of action of

addictive stimuli: Incentive-sensitization and addiction. Addiction, 96,

103–114.

Roefs, A., & Jansen, A. (2002). Implicit and explicit attitudes toward high-

fat foods in obesity. Journal of Abnormal Psychology, 111, 517–521.

Sherman, S. J., Rose, J. S., Koch, K., Presson, C. C., & Chassin, L. (2003).

Implicit and explicit attitudes toward cigarette smoking: The effects of

context and motivation. Journal of Social and Clinical Psychology, 22,

13–39.

Stice, E. (2002). Risk and maintenance factors for eating pathology: A

meta-analytic review. Psychological Bulletin, 128, 825–848.

Teachman, B. A., & Brownell, K. D. (2001). Implicit anti-fat bias among

health professionals: Is anyone immune? International Journal of

Obesity, 25, 1525–1531.

Wiers, R. W., van Woerden, N., Smulders, F. T. Y., & de Jong, P. J. (2002).

Implicit and explicit alcohol-related cognitions in heavy and light

drinkers. Journal of Abnormal Psychology, 111, 648–658.

Zack, M., Toneatto, T., & MacLeod, C. M. (1999). Implicit activation of

alcohol concepts by negative affective cues distinguishes between

problem drinkers with high and low psychiatric distress. Journal of

Abnormal Psychology, 108, 518–531.